It is well known to electrical engineers generally (and particularly to microwave engineers) that the frequency of an RF oscillator can be xe2x80x9cpulledxe2x80x9d (i.e. shifted from the frequency of oscillation which would be seen if the oscillator were coupled to an ideal impedance-matched pure resistance), if the oscillator sees an impedance which is different from the ideal matched impedance. Thus, a varying load impedance may cause the oscillator frequency to shift.1 
1Any electrical oscillator can be xe2x80x9cpulledxe2x80x9d to some extentxe2x80x94that is, its frequency can be shiftedxe2x80x94by changing the net impedance seen by the oscillator. However, in many systems which use oscillators, pulling of a resonant circuit""s frequency is undesirable. An oscillator which is too easily pulled may be overly susceptible to irrelevant external circumstances, such as changes in parasitic capacitance due to human proximity or temperature change. Normal techniques to avoid oscillator pulling include using isolation/buffering circuits between the oscillator and the variable load, and/or using a high-Q tuned circuit to stabilize the oscillator. 
The present application sets forth various innovative methods and systems which take advantage of this effect. In one class of embodiments, an unbuffered2 RF oscillator is loaded by an electromagnetic propagation structure which is electromagnetically coupled, by proximity, to a material for which real time monitoring is desired. The net complex impedance3 seen by the oscillator will vary as the
2An unbuffered oscillator is a oscillator without buffer amplifiers or attenuators. Amplifiers boost the output power and provide isolation from the load impedance changes. Attenuators decrease the amplitude while providing an isolation of two times the attenuation. In the load pulled oscillator configuration the oscillator feedback path that supplies the phase shift needed for oscillation is separated from the load. 
3A xe2x80x9ccomplexxe2x80x9d number is one which can be written as A+Bi, where A is the number""s xe2x80x9crealxe2x80x9d part, B is the number""s xe2x80x9cimaginaryxe2x80x9d part, and i2=xe2x88x921. These numbers are added according to the rule 
(A+Bi)+(C+Di)=(A+C)+(B+D)i, 
and are multiplied according to the rule
(A+Bi)(C+Di)=(ACxe2x88x92BD)+(AD+BC)i. 
Complex numbers are used in representing many electrical parameters. For example, impedance can be represented as a complex number whose real part is the resistance, and whose imaginary part is equal to the reactance (inductance or capacitance).
Similarly, permittivity can be represented as a complex number whose imaginary part represents resistive loss, and whose real part represents reactive loading, by the medium, of the propagating electromagnetic wave. characteristics of the material in the electromagnetic propagation structure varies. As this complex impedance changes, the oscillator frequency will vary. Thus, the frequency variation (which can easily be measured) can reflect changes in density (due to bonding changes, addition of additional molecular chains, etc.), ionic content, dielectric constant, or microwave loss characteristics of the medium under study. These changes will xe2x80x9cpullxe2x80x9d the resonant frequency of the oscillator system. Changes in the medium""s magnetic permeability will also tend to cause a frequency change, since the propagation of the RF energy is an electromagnetic process which is coupled to both electric fields and magnetic fields within the transmission line.
To help explain the use of the load-pull effect in the disclosed innovations, the electromagnetics of a dielectric-loaded transmission line will first be reviewed. If a transmission line is (electrically) loaded with a dielectric material, changes in the composition of the dielectric material may cause electrical changes in the properties of the line. In particular, the impedance of the line, and the phase velocity of wave propagation in the line, may change.
This can be most readily illustrated by first considering propagation of a plane wave in free space. The propagation of a time-harmonic plane wave (of frequency f) in a uniform material will satisfy the reduced wave equation
(∇2+k2)E=(∇2+k2)H=0, 
where
E is the electric field (vector),
H is the magnetic field (vector), and
∇2 represents the sum of second partial derivatives along the three spatial axes.
This equation can be solved to define the electric field vector E, at any point r and time t, as
E(r,t)=E0exp[i(kxc2x7rxe2x88x92xcfx89t)], 
where
k is a wave propagation vector which points in the direction of propagation and has a magnitude equal to the wave number k, and
xcfx89=Angular Frequency=2xcfx80f.
In a vacuum, the wave number k has a value xe2x80x9ck0xe2x80x9d which is
k0=xcfx89/c
=xcfx89(xcexc0xcex50)xc2xd,
where
xcexc0=Magnetic Permeability of vacuum (4xcfx80xc3x9710xe2x88x927 Henrys per meter),
xcex50=Electric Permittivity of vacuum ((1/36xcfx80)xc3x9710xe2x88x929 Farads per meter), and
c=Speed of light=(xcexc0xcex50)xe2x88x92xc2xd=2.998xc3x97108 meters/second.
However, in a dielectric material, the wave number k is not equal to k0; instead
k=xcfx89/(c(xcexcrxcex5r)xc2xd)
=xcfx89(xcexc0xcexcrxcex50xcex5r)xc2xd,
where
xcexcr=Relative Permeability of the material (normalized to the permeability xcexc0 of a vacuum), and
xcex5r=Relative Permittivity of the material (normalized to the permittivity xcex50 of a vacuum).
Thus, if the relative permeability xcexcr and/or the relative permittivity xcex5r vary, the wave number k and the wave propagation vector k will also vary, and this variation will typically affect the load pulled oscillator frequency.4 
4The full analysis of wave propagation in a cavity or at a boundary is much more complex, but in any case wave propagation will depend on the wave number, and the foregoing equations show how the wave number k can vary as the medium changes. See generally, e.g., R. Elliott, Electromagnetics (1966); J. Jackson, Classical Electrodynamics (2d ed. 1975); G. Tyras, Radiation and Propagation of Electromagnetic Waves (1969); R. Mittra and S. Lee, Analytical Techniques in the Theory of Guided Waves (1971); L. Lewin, Theory of Waveguides (1975); all of which are hereby incorporated by reference. 
In a typical free-running oscillator, the oscillator frequency is defined by a resonant feedback circuit (the xe2x80x9ctankxe2x80x9d circuit), and can also be pulled slightly by a reactive load,5 as noted above. Thus, such an oscillator can be broadly tuned by including a varactor in the tank circuit.6 
5The degree by which the reactive load can change the oscillator""s frequency will depend on the coupling coefficient between the load and the tank circuit. Thus, an increased coupling coefficient means that the oscillator frequency will be more sensitive to changes in the load element. However, the coupling coefficient should not be increased to the point where spectral breakup (multiple frequency operation) occurs, since this would render the desired measurement of the oscillator signal impossible. 
6This is one type of voltage-controlled oscillator (VCO). 
As the oscillator""s frequency is thus shifted, the phase difference between the energy incident on and reflected from the load element (which is preferably a shorted transmission line segment) will change. This phase difference will be equal to an exact multiple of 180xc2x0 at any frequency where the electrical length of the transmission line segment is an exact multiple of xcex/4.
As the oscillator frequency passes through such a frequency (i.e. one where the transmission line segment""s electrical length is equal to a multiple of xcex/4), the load""s net impedance will change from inductive to capacitive (or vice versa). As this occurs, the frequency of the oscillator may change abruptly rather than smoothly.7 This jump in frequency will be referred to as a frequency xe2x80x9chopxe2x80x9d.8 
7This change in frequency, as the load goes from capacitive (xe2x88x92180xc2x0) to inductive (+180xc2x0) or vice versa, is instantaneous if the equivalent parallel resistive part is large (e.g. greater than approximately 500 ohms in a 50 ohm system). 
8The amount by which the frequency shifts during the xe2x80x9chopxe2x80x9d will depend on the Q of the load element (as seen by the oscillator circuit), and on the coupling coefficient between the load element and the tank circuit. 
For a transmission line of length I which contains a dielectric material of relative dielectric constant xcex5r, the frequency at which one full wavelength (1xcex) exists in the transmission line is equal to c (the speed of light in vacuum, which is 2.995xc3x97108 meters/second) divided by the length of the line in meters and by the square root of the relative dielectric constant of the material:
Frequency1xcex=c/(lxcex5rxc2xd).
For example, for a one-foot-long line filled with a material having xcex5r=1, l=12 inches (=0.3048 meters), and
Frequency1xcex=(2.995xc3x97108)/(0.3048xc3x971.0)≈980 MHz.
However, since one wavelength actually contains two excursions from inductive to capacitive reactive impedances, only one-half wavelength is required to see one frequency hop of the load pulled oscillator. If the transmission line is terminated into a short or an open, the resulting effective length is increased to twice the actual length, since a standing wave is generated (due to the energy incident at the short or open being reflected back to the input of the transmission line). In essence, the energy travels down the line, gets reflected, and travels back to the input. With this taken into account, the first frequency with a wavelength long enough to cause a frequency xe2x80x9chopxe2x80x9d of the oscillator is one fourth the length calculated above, or 245 MHz.
Multiples of this first quarter-wavelength frequency will also cause the impedance seen at the input to the transmission line to go from inductive to capacitive reactance. The longer the transmission line, the greater the number of phase transitions that will occur. Longer line length also multiplies the phase changes that are brought about by a change in the dielectric constant. For every one-quarter wavelength change in the effective (electrical) length of the line, the complex impedance seen at the oscillator changes by 180xc2x0.
For example, suppose that a given oscillator, coupled into a low loss load with an electrical length of one-quarter wavelength (xcex/4), provides 50 MHz of load pulling frequency change (total excursion through all phases). If the monitored material changes enough to produce a change of only one degree of phase in the electrical length of the load, the oscillator frequency will change by 138.9 kHz. This represents an absolute resolution of 7.2xc3x9710xe2x88x926 degrees of phase change for each Hertz of sensitivity.9 For every additional quarter wavelength of line length, this sensitivity to phase is multiplied by 1.5. This is due to the change in phase being an additive function of every additional quarter wave in the measurement section.
9Even if the resolution of frequency measurement is only xc2x1100 Hz, this would still give an accuracy of better than one-thousandth of one degree. This is vastly better resolution than is possible with vector impedance systems (such as an HP 8510 Network Analyser). 
In a typical tuning frequency versus voltage plot for a VCO loaded into a shorted transmission line, the height of the xe2x80x9chopxe2x80x9d can be measured by holding the VCO tuning voltage constant, while a transmission line terminated into a short is varied in length10 to cause a full rotation of the impedance vector seen at the VCO""s input port. The resulting data of frequency versus length of the transmission line will show a jump in frequency (a delta frequency from the bottom of the xe2x80x9chopxe2x80x9d to the top of the xe2x80x9chopxe2x80x9d) which coincides with the delta frequency of the xe2x80x9chopxe2x80x9d seen when the VCO was swept using the tuning voltage.
10Such variable transmission lines are commonly used in the microwave industry, and are referred to as xe2x80x9cline stretchers.xe2x80x9d
Thus, if the VCO is swept across a frequency band and the number of frequency xe2x80x9chopsxe2x80x9d was counted, the number of xe2x80x9chopsxe2x80x9d reveals the number of wavelengths in the transmission line.11 
11More precisely, it will be found that the wavelengths at which hops are observed are separated from each other by one-quarter of the effective (electrical) length of the measurement section. 
This provides a means for determination of the range of dielectric constant change in a medium even when it rotates the phase vector multiple times (and therefore, the oscillator frequency returns to the same value multiple times). If the dielectric constant of the material in the transmission line is increased, then the above equations show that the frequency of the first full wavelength is decreased by the square root of the dielectric constant. Additionally, this means that the number of wavelengths at a fixed frequency increases with increasing dielectric constant. These facts imply that the VCO tuning curve will see more xe2x80x9chopsxe2x80x9d as the dielectric constant is increased due to the increasing fraction or whole wavelengths encountered.
Ideally, the oscillator will not cease oscillations (or break into multiple frequency oscillation or spectral breakup) into any load regardless of the load characteristics. However, this is not a strictly necessary condition for use of the disclosed method and system innovations.12 
12The second harmonic of the oscillator frequency is typically enhanced (becoming greater in amplitude than the fundamental frequency) just before the shift from inductive to capacitive impedance (or vice versa), due to the extreme non-linearities at this point. This does not hinder the use of load pulling as a measurement technique, since the measurement is typically made outside of this region of the impedance shift from inductive to capacitive. Alternatively, the second harmonic may be filtered out of the measurement. 
A measure of the dielectric loss of a material is typically given as the dielectric loss tangent (a unitless number) which is defined as the tangent of the imaginary part divided by the real part of the complex dielectric constant. Low loss materials are typically below a loss tangent equal to or less than 0.01. When the disclosed systems are used to measure materials with a high loss factor, the material""s absorption begins to dominate the load versus frequency effects, but a measurement capability still exists due to the sensitivity of the load pulling method.
A load-pull system also permits other information to be derived, regarding the substance being monitored.
Additional information can be obtained by retuning the VCO, so that the frequency is forced to change, and making another measurement at a much higher frequency. Since materials change properties versus frequency, the amount of frequency change due to load pulling will vary versus the frequency of operation.
A VCO will typically be designed to cover approximately one octave above its turn on frequency. If a VCO would not give enough frequency change to see the desired range of varying parameters versus operating frequency, an additional unbuffered oscillator, which runs at any frequency required to obtain appropriate data, may be switched into the coaxial line.
When two widely spaced frequencies are measured for a medium under study with a load pulled oscillator, the difference (delta) frequency between these two measurements will be unique for a given medium. This phenomena will aid in distinguishing constituents and the progress of mixing or reaction.
If the incident power and the reflected power is measured in a system where the final load is a short, the difference in powers will be twice the insertion loss of the medium (since two transits occur through the medium of interest). The insertion loss measurement will aid in determination of the changing conductivity of the medium or its change in absorption of the RF energy. This information can be related to the mixing or reaction products to further distinguish unique situations where the frequency change of the load pulled oscillator is not enough information or resolution by itself.
The magnetic permeability xcexcr can also be dynamically measured by the disclosed techniques. Since the velocity varies with (xcexcrxcex5r)xe2x88x92xc2xd, changes in xcexc4 will change the phase shift through a given physical length of line, and thus change the frequency of the oscillator.
A sample-containing waveguide, like that of the principally preferred embodiment, will typically have locations where the electric field is strong but the magnetic field is zero; at such locations only permittivity will affect the oscillator load pull frequency. However, there will also commonly be locations in a waveguide where the magnetic fields are locally strong and electric field is zero: at these locations, only the permeability will affect the propagation characteristics of the transmission line (and therefore contribute to the oscillator frequency).
A system can be built to sample (primarily) one of these parameters. For example, to sample the permeability, the coaxial transmission line will be terminated into a short where the medium of interest is located only in close proximity to the short. A waveguide structure supports very well defined electrical and magnetic field functions, and the sample can be suitably placed in such a structure to measure primarily the permeability.
Typical compounds and substances do not have varying magnetic permeabilities and therefore, most of the discussion will involve the changing complex permittivity. However, the effects of changing complex permeability will create similar changes in the oscillator load pulling characteristics. If a substance such as barium titanate is studied, the effect of the changing permeability must be considered along with the change in permittivity unless the system is designed specifically to measure only one of these.
An unusual feature of the oscillator configuration used with the present invention is the separation of the load of interest from the resonant circuit proper. The configuration used isolates the two through the active device. It is the non-linear behavior of the transistor that provides the changes in frequency as the load is changed. The loop gain of an oscillator must be unity with an appropriate phase shift to cancel the negative impedance""s imaginary part13 around the resonant loop. The initial gain of the active device must be greater than unity before oscillations can begin in order for the oscillator to be self starting. This extra gain is reduced to unity by the saturation of the active device upon establishment of the oscillations. Saturation of a device normally also changes the phase shift through the device 14. This requires a change in the operation frequency as the load changes due to the shift in loop gain and phase by the saturated condition change in the active device.
13In a simple resistor, an increase in the current passing through the resistor will produce an increase in the voltage across the resistor. By contrast, in microwave gain diodes (or in a transistor with feedback connections) which is operating at less than its saturated current, a small transient increase in the current across the device will produce a reduction in the voltage across the device. Thus, since a simple resistor has a positive impedance, such gain devices are referred to as having a negative impedance. 
14As the gain device approaches saturation, the physics of its operation will gradually change. These changes may cause the phase shift across the gain device to vary significantly. Note that, in the saturation regime, the gain device behaves as a non-linear circuit element. 
It has been discovered that, in a system using a free-running oscillator as described above, spectral purity of the oscillator is an important concern. Many microwave oscillators exhibit xe2x80x9cspectral breakup,xe2x80x9d wherein the spectrum of the oscillator""s output actually contains multiple frequencies. In most microwave oscillators this is not a problem, since a tuned feedback element will be used to stabilize the gain element, and/or isolation or buffering stages are used to prevent the oscillator""s feedback loop from being perturbed by extraneous resonances. However, in a load-pulled system, since such buffer stages are not used, spectral purity turns out to be quite important. For example, a spurious resonance in the feedback loop (e.g. due to a low-quality RF choke, or due to two impedance mismatches) can permit the oscillator to hop to a frequency which is determined (at least partly) by a harmonic of the spurious resonance, in which case the degree to which the oscillator frequency has been pulled by the changing load will be obscured.
To avoid such problems in a load-pulled system, a small series resistor can be interposed in the RF output of the oscillator, before the measurement section connection. This resistor adds a small amount of damping, which helps to suppress oscillation at secondary frequencies).
To further improve stability, a shunt resistor can be attached to the RF output of the load-pulled oscillator. This resistor adds to stability, by fixing a maximum magnitude for the load impedance seen at the RF output line.15 
15At frequencies where the length of the transmission line segment is a multiple of k/4), its impedance can become very large. 
Various types of apparatus have been proposed for measuring the concentration of one substance in another, particularly the concentration of a liquid or flowable substance in another liquid or flowable substance. Various devices which utilize the broad concept of determining composition of matter by measuring changes in a microwave signal are disclosed in U.S. Pat. Nos. 3,498,112 to Howard; 3,693,079 to Walker; 4,206,399 to Fitzky et al.; 4,311,957 to Hewitt et al.; 4,361,801 to Meyer et al.; 4,240,028 to Davis Jr.; 4,352,288 to Paap et al.; 4,499,418 to Helms et al.; and 4,367,440 and 4,429,273, both to Mazzagatti; all of which are hereby incorporated by reference.
Although various systems utilizing microwave transmissivity or signal alteration characteristics have been proposed in the prior art, certain considerations in utilizing microwave energy to detect the presence of the concentration of one medium in another have not been met by prior art apparatus. In particular, it is desirable in certain instances to be able to accurately measure, on a continuous basis, the concentration or change in concentration of one fluid in another and particularly where the concentration of one fluid is a very low percentage of the total fluid flow rate or fluid mixture quantity. It is also desirable that the signal change caused by the presence of one substance or medium in another be easily measured and be relatively error free, again, particularly in instances where measurements of low concentrations of one substance such as a fluid in another substance such as another fluid are being taken. Moreover, it is important to be able to transmit the microwave signal through a true cross section of the composition being sampled or measured to enhance the accuracy of the measurement.
Typical systems for capacitive based measurement have a capacitive element, used for parameter determination, as part of the resonant feedback loop around an active device. This method works well with very low loss systems, but oscillation ceases with even slightly lossy measurements. As the frequency is increased into the microwave region, it becomes difficult to configure the resonant feedback loop due to the increase in loss versus frequency and the wavelength becoming comparable to the path length. In this case the frequency is changed directly by the resonance change in the feedback loop which includes the element that consists of the sample to be measured. This frequency change is limited to the characteristics and loss of the feedback path and can only be changed over a narrow frequency range with out cessation of oscillations. This limits the measurement technique to small samples of very low loss.
At higher frequencies (above approximately 100 MHz), the capacitive measurement technique fails to work, due to line lengths and stray capacitances. At such frequencies resonant cavity techniques have been employed. (For example, a sample is placed in a resonant cavity to measure the loss and frequency shift with a external microwave frequency source that can be swept across the resonance with and without the sample in the cavity.) This method uses a highly isolated microwave frequency source which is forced by the user (rather than being pulled by the changing resonance) to change its frequency. This technique too meets substantial difficulties. For example, the use of multiple interfaces without a microwave impedance match at each interface causes extraneous reflections, which tend to hide the desired measurement data. This technique too gives errors with very lossy material, but in this case it is due to the very rounded nature of the resonance curve (which is due to the low Q of the loaded cavity). This rounded curve makes it difficult to determine both the center frequency and the 3 dB rolloff frequency closely enough to be accurate in the measurement.
Another technique which is used encompasses the use of a very sharp rise time pulse to obtain time domain data, from which frequency domain values are then derived through transformation techniques.
In U.S. Pat. No. 4,396,062 to Iskander, entitled Apparatus and Method for Time-Domain Tracking of High-speed Chemical Reactions, the technique used is time domain reflectometry (TDR). This contains a feedback system comprising a measurement of the complex permittivity by TDR means which then forces a change in frequency of the source which is heating the formation to optimize this operation. Additionally it covers the measurement of the complex permittivity by TDR methods.
U.S. Pat. No. 3,965,416 to Friedman appears to teach the use of pulse drivers to excite unstable, bi-stable, or relaxation circuits, and thereby propagate a pulsed signal down a transmission line which contains the medium of interest. The pulse delay is indicative of the dielectric constant of the medium. As in all cases, these are either square wave pulses about zero or positive or negative pulses. The circuit is a pulse delay oscillator where the frequency determining element is a shorted transmission line. The frequency generated is promoted and sustained by the return reflection of each pulse. The circuit will not sustain itself into a load that is lossy, since the re-triggering will not occur without a return signal of sufficient magnitude. In addition, the circuit requires a load which is a DC short in order to complete the DC return path that is required for re-triggering the tunnel diodes.
The frequencies of operation of any pulse system can be represented as a Fourier Series with a maximum frequency which is inversely dependent upon the rise time of the pulse. Therefore, the system covered in the Friedman patent is dependent upon the summation of the frequency response across a wide bandwidth. This causes increased distortion of the return pulse and prevents a selective identification of the dielectric constant versus frequency. This also forces a design of the transmission system to meet stringent criteria to prevent additional reflections across a large bandwidth.
The low frequency limit of the TDR technique is determined by the time window which is a function of the length of the transmission line. The upper extreme is determined by the frequency content of the applied pulse. In the case of this pulse delay line oscillator, the upper frequency is determined to a greater extent by the quality of impedance match (the lack of extra reflections) from the circuit through to the substance under study. These extra reflections would more easily upset the re-triggering at higher frequencies.
In one case (FIG. 1 of Friedman) the return reflection initiates a new pulse from the tunnel diode and therefore sets up a frequency (pulse repetition rate) as new pulses continue to be propagated. This is in essence a monostable multivibrator with the return reflection being the trigger. The problem implied, but not completely covered with this approach, is that due to the delay in pulses, the pulse train can overlap and cause multiple triggers to occur. These are caused by the re-reflections of the original parent pulse. An additional problem is with very lossy dielectrics, which will not provide enough feedback signal to initiate the next pulse. If the dielectric medium is of high enough dielectric constant to contain more than one wavelength, or if the dielectric constant of the samples vary greatly, multiple return reflections will alter the behavior of the circuit to render it useless due to the interfering train of return and parent pulses.
FIG. 3 of Friedman shows a bistable multivibrator which senses the return pulse by sampling and feeding back enough phase shifted voltage to re-set the tunnel diodes. Since this device is also dependent upon the return to trigger or re-trigger the parent pulse, it suffers problems with lossy dielectrics and high dielectric constant mediums.
To overcome these problems, the relaxation oscillator of FIG. 4 of Friedman was proposed that contains a RC (resistor/capacitor timing) network which will maintain the generation of pulse trains using resistor 76 and capacitor 78 with the dielectric filled transmission line affecting the regeneration of the pulses as the reflected parent pulse voltage is returned. Since the RC time constant is defining the basic repetition rate, some improvement is obtained in reducing second order effects. The transmission line is still an integral part of the overall relaxation oscillator and lossy dielectrics may cause irregular circuit response. The proposed inverting amplifier as the pulse generator will not function at above approximately 1 MHz in frequency due to the characteristics of such inverting amplifiers. The tunnel diode can pulse up to a 100 MHz rate.
By contrast, the innovative system embodiments disclosed in the present application and its parents differ from the known prior art in using a microwave frequency generated by a free running sine wave oscillator. The preferred oscillator has the versatile capability to work into a wide variety of transmission lines or other load impedance without generation of spurious data or cessation of oscillations. It will continue to oscillate with very lossy dielectrics. It is not a relaxation oscillator or a multivibrator. The frequency of the un-isolated oscillator is dependent upon the net complex impedance of the transmission line and will work into an open circuit as well as a short circuit. The net complex impedance at the frequency of operation of the oscillator looking at the transmission line containing the medium of interest results in stable oscillations through pulling of the unisolated oscillator. Only one frequency at any one time is involved in the disclosed system proposed (not counting harmonics which are at least 10 dB down from the fundamental). This provides for well defined information and eases the transmission design criteria. This also provides for evaluation of the dielectric constant versus frequency which can improve resolution of constituents or ionic activity.
Another important difference from prior art is the separation of the load of interest from the resonant circuit proper. The configuration used isolates the two through the transistor. It is the non-linear behavior of the transistor that provides the changes in frequency as the load is changed. The loop gain of an oscillator must be unity with 180xc2x0 phase shift. The initial gain of the transistor must be greater before oscillations begin in order for the oscillator to be self starting. This extra gain is reduced to unity by the saturation of the active device upon establishment of the oscillatory frequency. Saturating a device changes the gain (and accordingly the phase since it is non-linear) to maintain oscillations as the load changes. This will continue as the load changes as long as the transistor has appropriate phase and available gain to satisfy oscillations.
The use of aluminum oxide for moisture adsorption is well known in the industry. The surface attracts and retains water molecules by association with the bonds. Since this is a weak attraction there is a point at which the absorption and desorption reaches an equilibrium with the surrounding moisture content. Moisture measurements have been made with capacitance measurements using a very thin aluminum oxide surface with imbedded electrodes. When the water is absorbed the capacitance changes and therefore a measurement is made. This surface must be thin in order to allow the water molecules to accumulate in a region where the electrical field is present.
The present application discloses processes for monitoring the state of processing of, and for partially characterizing the composition of, food and feed products, by observing the frequency of a load-pull oscillator which is RF-coupled to the material under test (preferably by a simple single-ended RF probe).
Conventional process control in the food industries is almost entirely off-line (using laboratories to test samples). On-line controls are ordinarily limited to temperature, flow meters, viscosity, and mass (weighing systems). Processes are usually xe2x80x9crecipesxe2x80x9d of weights, times, and temperatures. This is because foodstuffs are chemically very complex, so that conventional high-tech methods (such as chromatographs and near IR spectroscopy) are not usually as adaptable to on line process control as in the xe2x80x9cregular chemicalxe2x80x9d industry. The materials are molecularly too complex.
Therefore, laboratory analysis must be used to measure (or infer) a process condition. Items as carbohydrate, fats, protein, fiber content, ash content, mineral content are done by conventional xe2x80x9cwetxe2x80x9d analysis. These lab methods are usually
1. Refractometry
2. Photoelectric Colorimetry
3. Some spectrophotometry
4. Some polarimetry
5. Melting/softening points
6. Viscometry
7. Conductivity
8. Some Chromatography
9. Titrations
10. Mass/Loss gravimetric methods
11. Solvent extractions techniques
Sometimes these are indirect measurements. For example, viscosity may be used to infer water content or gelatinization of starch (cooking). Water content of various components is a major item of interest/control. This is usually measured by heating a sample and measuring the weight loss. Color is used to determine proper cooking times for caramelization of starch/flour products.
By contrast, the disclosed methods permit direct real-time measurement of the molecular changes.
In one aspect of this, melting and softening can be measured directly, and correlation with temperature will then give an indirect measurements of process states.
Given a generally known process flow, the present invention provides new methods for monitoring the composition of the flow. For example, the disclosed inventions permit real-time non-contaminating measurement of water content, or fat content, or both in a stream of ingredients or in a stream of processed food products.
Given a generally known process flow, the present invention also provides new methods for monitoring the degree of cooking of the flow. As the following results show, the molecular changes in starches which are caused by cooking can be directly detected, and the molecular changes in meats which are caused by cooking can also be directly detected. This provides efficient endpoint detection for food processing.
The simplest way to use this monitoring technique analytically is to look at the time derivative of the measured RF frequencies: a certain percentage decrease in the rate of change can be used for an endpoint signal, to terminate a batch cooking stage. (Of course, this percentage decrease would be customized for a particular process, and would allow for continued cooking as the temperature of the food materials is ramped down.)
Among the disclosed inventions is provided a method for processing food and analogous materials, comprising the steps of: providing multiple flows of ingredient materials; electromagnetically coupling a single-ended RF probe to at least one the flow of ingredient materials, the probe being electrically connected to load a free-running RF oscillator, with no RF buffer stage being interposed between the oscillator and the probe; and observing the frequency behavior of the oscillator, to detect variation in the composition of the respective flow of ingredient materials; dynamically controlling the flows of the materials in accordance with results of the observing step; and combining and processing the flows of ingredient materials to provide a food product.
Among the disclosed inventions is provided a method for drying organic materials, comprising the steps of: providing a flow of a material which varies in water content; electromagnetically coupling a RF probe to the flow, the probe being electrically connected to load a free-running RF oscillator, with no RF buffer stage being interposed between the oscillator and the probe; observing the frequency behavior of the oscillator, to detect the moisture content of the flow; and adding water to the flow whenever the observing step indicates that the moisture content of the flow is below a target level; and drying the flow in a dryer stage; whereby the moisture content of the flow is dynamically controlled to be high enough to prevent clogging of the dryer, but no higher than necessary for reliable operation.
Among the disclosed inventions is provided a method for cooking food and analogous materials, comprising the steps of: introducing a mixture of predetermined ingredients into a cooking vessel; applying heat to the vessel in a controlled temperature-versus-time relationship, to cook the mixture; electromagnetically coupling a RF probe to the mixture in the vessel, and connecting the probe to load an oscillator operating at more than 100 MHz, with no RF buffer stage being interposed between the oscillator and the probe; observing the frequency behavior of the oscillator, to detect changes in the molecular composition and/or conformation of the mixture; and unloading the vat at a time which is at least partially determined by the results of the observing step.
Among the disclosed inventions is provided a methods for monitoring the state of processing of, and for partially characterizing the composition of, food and feed products, by observing the frequency of a load-pull oscillator which is RF-coupled to the material under test (preferably by a simple single-ended RF probe).
In many applications the avoidance of direct contact with the materials under test is overwhelmingly desirable, to prevent contamination.
The present application discloses a noninvasive RF probe which can be readily coupled, through a dielectric window, to a material under test. This probe provides a xe2x80x9csingle-endedxe2x80x9d isolated-coupling element which permits load-pull measurements to be made on an increased variety of materials. The electrical configuration of this probe is like that of a patch antenna,16 and hence this probe may be referred to as a xe2x80x9cpatch probexe2x80x9d. The patch probe is inherently less sensitive than a probe which is directly immersed in or inserted into the material under test, but may be sufficiently sensitive for many applications.
16Microwave antennas of such configuration are believed to have been used for heating, but apparently not for characterization. 
A planar probe can also be used for coupling through a window. In this case the planar probe would be placed flat against the window. However, the patch probe is preferred for such applications.
Among the disclosed inventions is provided a system for detecting the composition and microstructure of materials, comprising: an oscillator, which includes a gain element capable of providing substantial gain at frequencies greater than 100 MHz, and a feedback path, coupling an output of the gain element to an input thereof, the feedback path including a tunable resonant circuit; and a patch antenna which is RF-coupled to load the oscillator and which is placed in proximity to a portion of the material to be characterized so that electromagnetic wave propagation in the antenna is electrically loaded thereby; and circuitry connected to monitor the frequency of the oscillator to ascertain changes in the composition or microstructure of the material.
Among the disclosed inventions is provided a method for detecting the composition and microstructure of materials, comprising the steps of: providing a free-running oscillator which is connected to be pulled by the varying susceptance seen at a load connection thereto; connecting the load connection to the material under test through a patch antenna which is RF-coupled to load the oscillator and which is placed in proximity to a portion of the material to be characterized so that electromagnetic wave propagation in the antenna is electrically loaded thereby; and observing changes in the frequency of the oscillator.
Among the disclosed inventions is provided an electrical characterization system in which a patch antenna, used as a single-ended RF probe, is placed sufficiently close to a material under test to achieve near-field coupling thereto, and is also electrically coupled to a load-pulled oscillator.
The present application discloses a method for rapidly analyzing the state of a given process. A load-pulled oscillator is coupled to the material under test, and is swept across a range of frequencies. The oscillator frequency is swept, for example, by sweeping a tuning voltage, applied to a varactor in the oscillator circuits, across a predetermined range. The oscillator is coupled to the material under test by a probe which is electrically long (preferably at least several half-wavelengths when fully loaded by the material under test). The specific conditions (probe type, physical conditions of coupling, and range of tuning voltages or frequencies) will all have been previously defined, using the various considerations set forth in detail below. The oscillator frequency is monitored while the tuning voltage is swept in a predetermined direction (up rather than down, for example.
For this defined set of conditions, each sweep of the tuning voltage Vtun will produce a corresponding range of oscillator frequency values fosc. By integrating fosc over the predetermined range of Vtun, a single derived index number results. This turns out to be very useful in characterizing a given process under a given set of conditions.
Part of the reason for this is that shifts in material composition which produce even very small shifts in permittivity will have the effect of shifting the xe2x80x9ckneesxe2x80x9d in the frequency curve. These knees, which are readily visible in plots of oscillator frequency as a function of tuning voltage, correspond to points where the oscillator phase goes through a 180xc2x0 transition. When this occurs, the oscillator will return to its original operating frequency, and this frequency is likely to shift.
In a typical application the oscillator""s basic frequency can be forced to change by the inclusion of a varactor (a voltage variable capacitor) in the primary resonant loop of the circuit. By applying a DC voltage on this varactor, many oscillators can be tuned over an octave band. In the description above, the oscillator and load pull performance was assuming a fixed frequency (no varactor) circuit. If the load was a fixed length of lossless transmission line and the oscillator frequency was forced by the applied voltage on the varactor as opposed to the load pull phenomena, the xe2x80x9ckneesxe2x80x9d would be seen as the phase seen at the oscillator was swept through 180xc2x0 because of the effect of decreased wavelengths at higher frequencies. The number of knees appearing in the voltage vs frequency plot is dependent upon the dielectric constant of the medium in the transmission line, the length of the line and the frequency.
Thus, simple data reduction can be performed to derive a single index number for a given set of conditions. This is particularly useful where a given system is being tracked over time, since the time-domain behavior of the index number can easily be tracked. Thus, for instance, for endpoint detection in monitoring a batch process, the endpoint can be identified when the index value has shifted by a certain percentage from its initial value and the rate of change has declined to a certain percentage of its maximum value during the process run.
Among the disclosed inventions is provided a method for controlling a process, comprising the steps of: providing a voltage-controlled oscillator which is connected to be pulled by the varying susceptance seen at a load connection thereto, and which is connected to be tuned by a tuning voltage applied thereto; connecting the load connection to an RF interface which is electrically loaded by proximity to material undergoing the process; sweeping the tuning voltage across a predetermined range of voltages; integrating the oscillation frequency of the oscillator, as a function of tuning voltage, across the range of voltages, to provide a process index value; comparing the process index value with a known range of values for comparable process conditions; and taking action conditionally, within the process, in dependence on the result of the comparing step.
Among the disclosed inventions is provided a process control method, wherein a load-pulled voltage-controlled oscillator is coupled through an RF probe, without isolation, to a material in the process. The frequency response of the oscillator is then integrated over voltage, as the tuning voltage is varied across a predetermined range. This integral gives a single xe2x80x9cprocess indexxe2x80x9d value which is then used as a basis for conditional action on the process.
A further disclosed innovation is a single-ended probe which includes multiple transmission line segments, and which also includes an RF switching element connected to permit switching between the two segments.
If an RF switch (pin diodes) was used on the substrate to switch between two lines, one could be an uncovered metal trace and the other could be a covered metal section with the covering being selective to a particular chemical. This combination would provide a measurement of a specific substance using the covered side of the probe, and once this component of the material under study is known an additional component could be derived from the response from the bare side of the probe. For example if the covered side was of the material to discern glucose in a dextrose/glucose/water mixture, the bare side""s additional information would provide for a solution to how much water was in the mixture.
This can also be used to provide spatially-resolved differential measurement for detection of spatially-varying characteristics (e.g. material zone boundaries in a distillation or chromatographic column).
Among the disclosed inventions is provided a system for detecting the composition and microstructure of materials, comprising: an oscillator, which includes a gain element capable of providing substantial gain at frequencies greater than 100 MHz, and a feedback path, coupling an output of the gain element to an input thereof, the feedback path including a tunable resonant circuit; and an electromagnetic propagation structure which is RF-coupled to load the oscillator and which includes an RF switch and first and second transmission line structures, the switch being connected and configured to connect the first transmission line structure to the external connection selectively under remote command; at least one of the transmission line structures being positioned so that electromagnetic wave propagation thereon can be electrically loaded by proximity to a portion of the material to be characterized; and circuitry connected to monitor the frequency of the oscillator to ascertain changes in the composition or microstructure of the material.
Among the disclosed inventions is provided a single-ended RF probe, for providing a bidirectional RF interface to materials to be characterized, comprising: an external RF connection mechanically connected to a dielectric support structure; and an RF switch mounted on the support structure and electrically connected to the external connection; and first and second transmission line structures, each connected to the switch and mounted on the support structure; wherein the switch is connected and configured to connect the first transmission line structure to the external connection selectively, in accordance with a bias signal received at the external connection.
Among the disclosed inventions is provided a method for detecting the composition and microstructure of materials, comprising the steps of: providing a tunable oscillator which is connected to be pulled by the varying susceptance seen at a load connection thereto; connecting the load connection to the material under test through a single-ended probe which includes an RF switch and first and second transmission line structures, the switch being connected and configured to connect the first transmission line structure to the external connection selectively under remote command; positioning the probe so that at least one of the transmission line structures is electrically loaded by proximity to a portion of the material to be characterized; and observing changes in the frequency of the oscillator, while switching the RF switch to activate the first and second transmission lines alternately.
Among the disclosed inventions is provided a single-ended RF probe which contains an RF switch, and TWO transmission lines, all integrated into a common mechanical structure. The two transmission lines can both be capacitively loaded by inserting the mechanical structure into a material under test, but the two lines have different coupling characteristics. (For example, one line may be coated with a selective absorption material; or the two lines may merely be spatially separate.) Remote sensing electronics, such as a load-pulled oscillator, thus have an electrical interface through which to detect changes corresponding to the properties of the material under test. The RF switch permits additional information to be gained by switching between the two transmission lines.
It is hereby disclosed to form a xe2x80x9cbedxe2x80x9d of an absorbent media between a coaxial cable consisting of a center conductor and an enclosing metallic mesh. The mesh allows a stream of liquid or gas to pass through the structure, so that the media will adsorb the material to which it is specific. This changes the permittivity of the media in which the electromagnetic field is propagating. This change in permittivity can be seen through the use of classical microwave methods such as phase shift, amplitude changes, frequency changes in a cavity or the frequency of a unbuffered oscillator.
In at least some embodiments, it is disclosed to use a two cylinder structure, where an outer cylinder contains a material which selectively removes a chemical which may be in conflict with or would contaminate the sensing of the desired chemical. This outer cylinder does not play a part in the measurement because it is outside the metal shield which contains the measurement adsorbent, and is thus outside the electromagnetic field.
In at least some embodiments, it is disclosed to use two different types of adsorbent in the main measurement portion, each having somewhat different properties with regard to the chemical to be measured.
The disclosed innovations, in various embodiments, provide one or more of at least the following advantages:
greater sensitivity to low concentrations than in prior methods;
apparatus is not permanently altered by lossy materials or by material which coat the substrate, as are capacitance-based measurements.